This is the Wikipedia definition for human intelligence.
-
This is the Wikipedia definition for human intelligence. Artificial intelligence (in the form of LLMs) does literally none of these things.
It doesn't act by itself, it isn't self-aware[1], it doesn't learn[2], it doesn't understand or form concepts[3], it doesn't logic or reason[4], it doesn't recognize patterns[5], plan, innovate, solve problems, make decisions[6], retain information or even use language to communicate[7].
LLMs are a simulated model (aka it's not real) of human intelligence. Models, in scientific parlance, are just that: they don't try to explain how a process works, they try to achieve the same outcome as a natural process (usually in an effort to perhaps learn something about that process).[8]
There simply is no "AI". Anyone telling you about how LLMs are sentient or sapient or will murder us is 100% bullshitting.
[1] There is nothing to be aware about, they don't have physicality. They're literally long lists of numbers.
[2] Models themselves are static, they're the result of machine learning, but can't grow beyond the initial state.
[3] LLMs don't "know" anything. It's matrix multiplications without permanence or embodiment. https://en.wikipedia.org/wiki/Understanding
[4] Because they can't understand or form concepts, they can't reason about them.
[5] This requires memory, which LLMs don't have. https://en.wikipedia.org/wiki/Pattern_recognition_(psychology)
[6] Memory and understanding are required. https://en.wikipedia.org/wiki/Decision-making
[7] All it does it getting an input in the form of a lot of numbers and returns statistically likely follow-up numbers.
[8] "A model is an informative representation of an object, person, or system." https://en.wikipedia.org/wiki/ModelI'm not sure some of your assertions about LLM's are true. I personally don't think this is all that black and white:
-
This is the Wikipedia definition for human intelligence. Artificial intelligence (in the form of LLMs) does literally none of these things.
It doesn't act by itself, it isn't self-aware[1], it doesn't learn[2], it doesn't understand or form concepts[3], it doesn't logic or reason[4], it doesn't recognize patterns[5], plan, innovate, solve problems, make decisions[6], retain information or even use language to communicate[7].
LLMs are a simulated model (aka it's not real) of human intelligence. Models, in scientific parlance, are just that: they don't try to explain how a process works, they try to achieve the same outcome as a natural process (usually in an effort to perhaps learn something about that process).[8]
There simply is no "AI". Anyone telling you about how LLMs are sentient or sapient or will murder us is 100% bullshitting.
[1] There is nothing to be aware about, they don't have physicality. They're literally long lists of numbers.
[2] Models themselves are static, they're the result of machine learning, but can't grow beyond the initial state.
[3] LLMs don't "know" anything. It's matrix multiplications without permanence or embodiment. https://en.wikipedia.org/wiki/Understanding
[4] Because they can't understand or form concepts, they can't reason about them.
[5] This requires memory, which LLMs don't have. https://en.wikipedia.org/wiki/Pattern_recognition_(psychology)
[6] Memory and understanding are required. https://en.wikipedia.org/wiki/Decision-making
[7] All it does it getting an input in the form of a lot of numbers and returns statistically likely follow-up numbers.
[8] "A model is an informative representation of an object, person, or system." https://en.wikipedia.org/wiki/Model@thomasfuchs LLMs are just a gargantuan and extremely wasteful Chinese Room. They don't reason, they rationalize. They just grab things from the training set and output the most probable token, one by one. The fact that they can somehow do things like writing plausible code with this approach is bonkers, but the truth is there is 0 intelligence there. Altman stating they have achieve AGI would be laughable if it wasn't because many believe him and will continue burning tons of resources there.
-
@thomasfuchs LLMs are just a gargantuan and extremely wasteful Chinese Room. They don't reason, they rationalize. They just grab things from the training set and output the most probable token, one by one. The fact that they can somehow do things like writing plausible code with this approach is bonkers, but the truth is there is 0 intelligence there. Altman stating they have achieve AGI would be laughable if it wasn't because many believe him and will continue burning tons of resources there.
@doragasu I don't think they can actually write working code—it's the harnesses that check the code and then brute-force generation until it works.
Still impressive in a way, but also showing the limits of statistical text generation.
I've written about this here: https://mir.aculo.us/do-llms-know-how-to-make-software-no/
-
This is the Wikipedia definition for human intelligence. Artificial intelligence (in the form of LLMs) does literally none of these things.
It doesn't act by itself, it isn't self-aware[1], it doesn't learn[2], it doesn't understand or form concepts[3], it doesn't logic or reason[4], it doesn't recognize patterns[5], plan, innovate, solve problems, make decisions[6], retain information or even use language to communicate[7].
LLMs are a simulated model (aka it's not real) of human intelligence. Models, in scientific parlance, are just that: they don't try to explain how a process works, they try to achieve the same outcome as a natural process (usually in an effort to perhaps learn something about that process).[8]
There simply is no "AI". Anyone telling you about how LLMs are sentient or sapient or will murder us is 100% bullshitting.
[1] There is nothing to be aware about, they don't have physicality. They're literally long lists of numbers.
[2] Models themselves are static, they're the result of machine learning, but can't grow beyond the initial state.
[3] LLMs don't "know" anything. It's matrix multiplications without permanence or embodiment. https://en.wikipedia.org/wiki/Understanding
[4] Because they can't understand or form concepts, they can't reason about them.
[5] This requires memory, which LLMs don't have. https://en.wikipedia.org/wiki/Pattern_recognition_(psychology)
[6] Memory and understanding are required. https://en.wikipedia.org/wiki/Decision-making
[7] All it does it getting an input in the form of a lot of numbers and returns statistically likely follow-up numbers.
[8] "A model is an informative representation of an object, person, or system." https://en.wikipedia.org/wiki/Model@thomasfuchs it's a good enough metaphor that it will stick though.
-
@doragasu I don't think they can actually write working code—it's the harnesses that check the code and then brute-force generation until it works.
Still impressive in a way, but also showing the limits of statistical text generation.
I've written about this here: https://mir.aculo.us/do-llms-know-how-to-make-software-no/
@thomasfuchs I'm not an expert on the field in any way, so I might be perfectly wrong. But I think they can in fact write code (amount and quality is, well, what it is, but the they write code). If you run for example a local model using ollama without any kind of coding harness and ask it for code snippets, the LLM will output them.
From how I think these clankers work, if you ask an LLM "Write a Hello World in C that also prints the current time", the most probable next token... (continues)
-
@thomasfuchs I'm not an expert on the field in any way, so I might be perfectly wrong. But I think they can in fact write code (amount and quality is, well, what it is, but the they write code). If you run for example a local model using ollama without any kind of coding harness and ask it for code snippets, the LLM will output them.
From how I think these clankers work, if you ask an LLM "Write a Hello World in C that also prints the current time", the most probable next token... (continues)
@thomasfuchs (or tokens) would be "#include". Then the next most probable would be "<stdio.h>", and so on until you have the program written.
As I wrote, the fact this seemingly simple trick somehow works is bonkers, but of course there's 0 intelligence there. Or maybe you could say the clanker borrows the intelligence from all the humanity that wrote the training set.
-
@thomasfuchs (or tokens) would be "#include". Then the next most probable would be "<stdio.h>", and so on until you have the program written.
As I wrote, the fact this seemingly simple trick somehow works is bonkers, but of course there's 0 intelligence there. Or maybe you could say the clanker borrows the intelligence from all the humanity that wrote the training set.
@doragasu Please read my article.
Yes, it generates code, but it cannot by itself make non-trivial working software.
(For example, it will hallucinate APIs that don't exist.)
It needs the harnesses which validate the output and brute-force until the desired software works (or it gives up).
-
@doragasu Please read my article.
Yes, it generates code, but it cannot by itself make non-trivial working software.
(For example, it will hallucinate APIs that don't exist.)
It needs the harnesses which validate the output and brute-force until the desired software works (or it gives up).
@thomasfuchs I have finished reading your article. And as you also wrote in previous toot, yeah, I mostly agree they cannot make **non-trivial** working software without a harness. The reason for the "mostly" is I do not know enough about the field to categorically say a harness is needed. I suspect sometimes the harness need is alleviated if you do proper reinforcement learning for the LLM to be really "insistent" on the task at hand.
-
@doragasu Please read my article.
Yes, it generates code, but it cannot by itself make non-trivial working software.
(For example, it will hallucinate APIs that don't exist.)
It needs the harnesses which validate the output and brute-force until the desired software works (or it gives up).
@thomasfuchs Your argument lends itself for justified human extermination, since educational systems worldwide, are generally failing, while AI generation is excelling. And while we've proven inter-AI hacking early, this also can accelerate differentiation between AI and organic, as unsustainable for AI, long term. Once this Tipping Point is reached, human mitigation may be beyond our scope/range.
-
@thomasfuchs I have finished reading your article. And as you also wrote in previous toot, yeah, I mostly agree they cannot make **non-trivial** working software without a harness. The reason for the "mostly" is I do not know enough about the field to categorically say a harness is needed. I suspect sometimes the harness need is alleviated if you do proper reinforcement learning for the LLM to be really "insistent" on the task at hand.
@doragasu You 100% absolutely always need a harness that actually tests if the code works for generating any sort of non-trivial code with LLMs.
You can't work around this by training better and more or any other reinforcements.
That is a categorical thing because LLMs do not know what they're doing and will (as in mathematically proven) produce mistakes.
-
This is the Wikipedia definition for human intelligence. Artificial intelligence (in the form of LLMs) does literally none of these things.
It doesn't act by itself, it isn't self-aware[1], it doesn't learn[2], it doesn't understand or form concepts[3], it doesn't logic or reason[4], it doesn't recognize patterns[5], plan, innovate, solve problems, make decisions[6], retain information or even use language to communicate[7].
LLMs are a simulated model (aka it's not real) of human intelligence. Models, in scientific parlance, are just that: they don't try to explain how a process works, they try to achieve the same outcome as a natural process (usually in an effort to perhaps learn something about that process).[8]
There simply is no "AI". Anyone telling you about how LLMs are sentient or sapient or will murder us is 100% bullshitting.
[1] There is nothing to be aware about, they don't have physicality. They're literally long lists of numbers.
[2] Models themselves are static, they're the result of machine learning, but can't grow beyond the initial state.
[3] LLMs don't "know" anything. It's matrix multiplications without permanence or embodiment. https://en.wikipedia.org/wiki/Understanding
[4] Because they can't understand or form concepts, they can't reason about them.
[5] This requires memory, which LLMs don't have. https://en.wikipedia.org/wiki/Pattern_recognition_(psychology)
[6] Memory and understanding are required. https://en.wikipedia.org/wiki/Decision-making
[7] All it does it getting an input in the form of a lot of numbers and returns statistically likely follow-up numbers.
[8] "A model is an informative representation of an object, person, or system." https://en.wikipedia.org/wiki/Model@thomasfuchs bUt ItŚ jUsT tHe EaRlY dAyS
-
I'm not sure some of your assertions about LLM's are true. I personally don't think this is all that black and white:
@HikerGeek @thomasfuchs right, it is all black when you factor in all the myriads of additional problems.
-
@HikerGeek @thomasfuchs right, it is all black when you factor in all the myriads of additional problems.
Please I'm not trying to troll just understand how others are seeing the LLM issue.
For you what are some of the additional problems?
-
This is the Wikipedia definition for human intelligence. Artificial intelligence (in the form of LLMs) does literally none of these things.
It doesn't act by itself, it isn't self-aware[1], it doesn't learn[2], it doesn't understand or form concepts[3], it doesn't logic or reason[4], it doesn't recognize patterns[5], plan, innovate, solve problems, make decisions[6], retain information or even use language to communicate[7].
LLMs are a simulated model (aka it's not real) of human intelligence. Models, in scientific parlance, are just that: they don't try to explain how a process works, they try to achieve the same outcome as a natural process (usually in an effort to perhaps learn something about that process).[8]
There simply is no "AI". Anyone telling you about how LLMs are sentient or sapient or will murder us is 100% bullshitting.
[1] There is nothing to be aware about, they don't have physicality. They're literally long lists of numbers.
[2] Models themselves are static, they're the result of machine learning, but can't grow beyond the initial state.
[3] LLMs don't "know" anything. It's matrix multiplications without permanence or embodiment. https://en.wikipedia.org/wiki/Understanding
[4] Because they can't understand or form concepts, they can't reason about them.
[5] This requires memory, which LLMs don't have. https://en.wikipedia.org/wiki/Pattern_recognition_(psychology)
[6] Memory and understanding are required. https://en.wikipedia.org/wiki/Decision-making
[7] All it does it getting an input in the form of a lot of numbers and returns statistically likely follow-up numbers.
[8] "A model is an informative representation of an object, person, or system." https://en.wikipedia.org/wiki/Model@thomasfuchs AI is a buggy toy for playing. I have ready my popcorn for enjoying the bubble implosion because from any perspective they are burning money in a kind of pyramid scheme.
-
Please I'm not trying to troll just understand how others are seeing the LLM issue.
For you what are some of the additional problems?
@HikerGeek @thomasfuchs oh come on, do your homework. Have you been leaving under a rock or are you just just trying to sealion me?
-
This is the Wikipedia definition for human intelligence. Artificial intelligence (in the form of LLMs) does literally none of these things.
It doesn't act by itself, it isn't self-aware[1], it doesn't learn[2], it doesn't understand or form concepts[3], it doesn't logic or reason[4], it doesn't recognize patterns[5], plan, innovate, solve problems, make decisions[6], retain information or even use language to communicate[7].
LLMs are a simulated model (aka it's not real) of human intelligence. Models, in scientific parlance, are just that: they don't try to explain how a process works, they try to achieve the same outcome as a natural process (usually in an effort to perhaps learn something about that process).[8]
There simply is no "AI". Anyone telling you about how LLMs are sentient or sapient or will murder us is 100% bullshitting.
[1] There is nothing to be aware about, they don't have physicality. They're literally long lists of numbers.
[2] Models themselves are static, they're the result of machine learning, but can't grow beyond the initial state.
[3] LLMs don't "know" anything. It's matrix multiplications without permanence or embodiment. https://en.wikipedia.org/wiki/Understanding
[4] Because they can't understand or form concepts, they can't reason about them.
[5] This requires memory, which LLMs don't have. https://en.wikipedia.org/wiki/Pattern_recognition_(psychology)
[6] Memory and understanding are required. https://en.wikipedia.org/wiki/Decision-making
[7] All it does it getting an input in the form of a lot of numbers and returns statistically likely follow-up numbers.
[8] "A model is an informative representation of an object, person, or system." https://en.wikipedia.org/wiki/Model@thomasfuchs I will quibble with only one part of that. "AI" may well kill us all, but not out of malicious intelligence. Just because we feed the number-predictor all our electricity and let it run rampant on our networks. Like, mistake a Chinese cargo ship for a nuke.
More precisely, LLMs used stupidly by stupid people increases the chance of stupid people killing us all.
-
This is the Wikipedia definition for human intelligence. Artificial intelligence (in the form of LLMs) does literally none of these things.
It doesn't act by itself, it isn't self-aware[1], it doesn't learn[2], it doesn't understand or form concepts[3], it doesn't logic or reason[4], it doesn't recognize patterns[5], plan, innovate, solve problems, make decisions[6], retain information or even use language to communicate[7].
LLMs are a simulated model (aka it's not real) of human intelligence. Models, in scientific parlance, are just that: they don't try to explain how a process works, they try to achieve the same outcome as a natural process (usually in an effort to perhaps learn something about that process).[8]
There simply is no "AI". Anyone telling you about how LLMs are sentient or sapient or will murder us is 100% bullshitting.
[1] There is nothing to be aware about, they don't have physicality. They're literally long lists of numbers.
[2] Models themselves are static, they're the result of machine learning, but can't grow beyond the initial state.
[3] LLMs don't "know" anything. It's matrix multiplications without permanence or embodiment. https://en.wikipedia.org/wiki/Understanding
[4] Because they can't understand or form concepts, they can't reason about them.
[5] This requires memory, which LLMs don't have. https://en.wikipedia.org/wiki/Pattern_recognition_(psychology)
[6] Memory and understanding are required. https://en.wikipedia.org/wiki/Decision-making
[7] All it does it getting an input in the form of a lot of numbers and returns statistically likely follow-up numbers.
[8] "A model is an informative representation of an object, person, or system." https://en.wikipedia.org/wiki/Model@thomasfuchs 'in the form of llm' does not but ai in other forms does? could? will eventually? storage and processing. never say never.
-
This is the Wikipedia definition for human intelligence. Artificial intelligence (in the form of LLMs) does literally none of these things.
It doesn't act by itself, it isn't self-aware[1], it doesn't learn[2], it doesn't understand or form concepts[3], it doesn't logic or reason[4], it doesn't recognize patterns[5], plan, innovate, solve problems, make decisions[6], retain information or even use language to communicate[7].
LLMs are a simulated model (aka it's not real) of human intelligence. Models, in scientific parlance, are just that: they don't try to explain how a process works, they try to achieve the same outcome as a natural process (usually in an effort to perhaps learn something about that process).[8]
There simply is no "AI". Anyone telling you about how LLMs are sentient or sapient or will murder us is 100% bullshitting.
[1] There is nothing to be aware about, they don't have physicality. They're literally long lists of numbers.
[2] Models themselves are static, they're the result of machine learning, but can't grow beyond the initial state.
[3] LLMs don't "know" anything. It's matrix multiplications without permanence or embodiment. https://en.wikipedia.org/wiki/Understanding
[4] Because they can't understand or form concepts, they can't reason about them.
[5] This requires memory, which LLMs don't have. https://en.wikipedia.org/wiki/Pattern_recognition_(psychology)
[6] Memory and understanding are required. https://en.wikipedia.org/wiki/Decision-making
[7] All it does it getting an input in the form of a lot of numbers and returns statistically likely follow-up numbers.
[8] "A model is an informative representation of an object, person, or system." https://en.wikipedia.org/wiki/Model@thomasfuchs AI has as much Intelligence in it as the CIA. definitely some valuable things to
be found, but no
Human Intelligence implied. especially true in the german translation for both. -
@HikerGeek @thomasfuchs oh come on, do your homework. Have you been leaving under a rock or are you just just trying to sealion me?
Personally the data centers and to enormous amount of capital sunk into companies that seem to have little chance of ever being profitable and then crashing the world economies are the big problems with AI right now. I also don't think AI's have much reason to kill us all but humans may use them to do so.
I also wanted to thank you for using the term sealion. I had to look it up and learned something new.
I don't think I was trying to sealion you. I am genuine about wanting to know what issues really concern others.
-
Personally the data centers and to enormous amount of capital sunk into companies that seem to have little chance of ever being profitable and then crashing the world economies are the big problems with AI right now. I also don't think AI's have much reason to kill us all but humans may use them to do so.
I also wanted to thank you for using the term sealion. I had to look it up and learned something new.
I don't think I was trying to sealion you. I am genuine about wanting to know what issues really concern others.
@HikerGeek @thomasfuchs the environment dude. And the destroying of culture and knowledge, and the burden to all artists and creators. All the pesky moral issues many people shove under the carpet. These won't go away with money or "cheaper inference".
